gondamol/kenya-ewars

# Kenya Climate-Sensitive Disease Early Warning System (EWARS) An open-source, AI-backed Early Warning System for Malaria, Dengue, and Cholera in Kenya. This system integrates climate data (rainfall, temperature) with disease surveillance data to predict outbreaks and generate actionable Situation Reports (SitReps) for public health officials. ## 🚀 Key Features - **Multi-Source Ingestion:** CHIRPS (Rainfall), NASA POWER (Temp), WHO (Disease), OSM (Clinics). - **Advanced Analytics:** Anomaly detection (CUSUM), Forecasting (Prophet), and Risk Modeling (Random Forest). - **Automated Reporting:** Generates PDF SitReps and interactive dashboards. - **Free & Open:** Built entirely on open-source tools (DuckDB, dbt, Streamlit, Prefect). ## 🛠️ Architecture - **Data Lake:** DuckDB (local file-based). - **ETL:** Python + dbt-core. - **Orchestration:** Prefect. - **Dashboard:** Streamlit. ## 📦 Quick Start ### Prerequisites - Docker & Docker Compose - Python 3.10+ ### Run Locally (Docker) ```bash # 1. Clone the repo git clone github.com cd kenya-ewars # 2. Build and Run docker-compose up --build ``` Access the dashboard at `localhost`. ### Run Locally (Python) ```bash # 1. Create venv python -m venv venv source venv/bin/activate # 2. Install dependencies pip install -r requirements.txt # 3. Run Pipeline (Ingest -> Model) python -m src.prefect_flows.flows # 4. Launch Dashboard streamlit run src/dashboard/app.py ``` ## ⚠️ Ethics & Limitations - **Data Privacy:** This system uses aggregated county-level data. No PII is processed. - **Model Limitations:** Predictions are probabilistic. They should **augment** human decision-making, not replace it. - **Validation:** All alerts must be verified by local health officers before action. ## 🤝 Contributing We welcome contributions! Please see `docs/design.md` for architecture details. ## 📄 License MIT License.